Applications of Reinforcement Learning and Its Extension to Tactical Simulation Technologies
Arif Furkan Mendi, Dilara Doğan, Tolga Erol and Turan Topaloğlu (HAVELSAN, Turkey); M. Esat Kalfaoğlu and Hüseyin Oktay Altun (AutoDidactic Technologies, Turkey)
Reinforcement Learning (RL) is a branch of machine learning that is used in many areas, from robotics to natural language processing, from game technologies to medical fields and finance. It is widely used in systems containing large data, where instantaneous data flow is intense and where data tagging is arduous or impossible. RL is a preferred approach for exploring new strategies or combinations due to its convenient nature to real life control problems where sequential decision-making is crucial. In the early stages of its development, RL was mainly prevalent in game technologies. However, recently applications of RL have diversified and extended to a plethora of new fields. As HAVELSAN, the leading Turkish defence industry software and simulation company, we investing this technique to take the eye-catching advantages. In this study, we elaborate on the fundamentals of reinforcement learning technology with an emphasis of its novel applications and future projections in the light of existing research findings. We then cover the RL specifically from simulation technologies perspective and introduce the FIVE-ML project of HAVELSAN which aims a transition from rule-based behaviour modelling to learning-based smart behaviour modelling in order to provide more effective and more dynamic pilot training environment.
Journal: International Journal of Simulation- Systems, Science and Technology- IJSSST V22
Published: no date/time given